help > Several questions for non-parametric stats and post-hoc
May 16, 2018  12:05 AM | Hyunwoo Kim
Several questions for non-parametric stats and post-hoc
Hi, CONN experts.

I'm doing 3 groups comparison using CONN.
I did F-test by contrast [1 -1 0; 0 1 -1] and got some significant area.(seed-to-voxel analysis & 2nd level analysis).

Statistic settings are below as CONN default gives;
 - height threshold p<0.001(p-uncorrected), cluster threshold p<0.05(cluster-size p-FDR corrected)
 - non-parametric stats with permutations = 1000

1) How can I determine the number of permutation?
   Because I don't know much about statistics, I don't know how to decide the number of permutations.
When I changed p-uncorrected value from 0.001 to 0.0001, CONN gives permutation number as 10000 and significant result disappeared.
Is there relationship between uncorrected p and number of iteration?

2) Is post-hoc available in CONN? If so, how?
As I mentioned, I've got some significant result from F-test. So I want to know where the difference come from.
In my limited knowledge, post-hoc would give answer but I don't know how to do it. If it is available in CONN, even available not in GUI but in command, please show me the way.

Thank you.
Threaded View
Title
Author
Date
Several questions for non-parametric stats and post-hoc
May 16, 2018